Aluminum nitride (AlN) holds great promise as a semiconductor material for numerous applications, e.g., opto-electronic devices such as short-wavelength light-emitting diodes (LEDs) and lasers, dielectric layers in optical storage media, electronic substrates, and chip carriers where high thermal conductivity is essential, among many others. In principle, the properties of AlN may allow light emission down to around 200 nanometers (nm) wavelength to be achieved. The use of AlN substrates is also expected to improve high-power radio-frequency (rf) devices, made with nitride semiconductors, due to the high thermal conductivity with low electrical conductivity. Addressing various challenges can help increase the commercial practicability of such devices.
For example, large-diameter bulk AlN crystals (grown, for example, using the techniques described in U.S. application Ser. No. 11/503,660, incorporated herein in its entirety, referred to hereinafter as the “'660 application”), may in some circumstances grow with hexagonal-prism shaped cavities defects that are roughly 0.5 millimeter (mm) in diameter and 0.1 mm thick. Area concentrations as high as 100 cm−2 have been observed in AlN slices that are cut to be 0.5 mm thick from these large diameter boules. Similar kinds of defects have been observed in the growth of other hexagonal crystals, such as SiC, and are commonly referred to as planar defects. These defects may be problematic for the further development of nitride-based electronics. In particular, they typically cause the surface of a substrate to roughen when they intersect the surface plane. They may also scatter light, which may be problematic for many opto-electronic applications that benefit from the transparency of AlN substrates at optical wavelengths between 210 and 4500 nm. Planar defects may also reduce the thermal conductivity around the defect, an effect that is generally undesirable for high-power devices in which the high intrinsic thermal conductivity of the AlN is useful. They may also introduce small angle grain boundaries into the AlN crystal and, thus, degrade the quality of the crystal by increasing the effective concentration of dislocations that thread through from one side of the wafer to the other (so-called threading dislocations) and that degrade the quality of surface preparation. Thus, the application of AlN substrates to the fabrication of high-performance, high-power opto-electronic and electronic devices may be enhanced if planar defects are reduced or eliminated.
Generally, planar defect formation in crystals grown by physical vapor transport (PVT) is caused by voids that get trapped in the growing crystal and that move and are shaped by the thermal gradients to which the crystal is exposed. A common cause identified in SiC crystal growth is poor seed attachment, where any kind of a microscopic void will commonly result in the formation of a planar defect (see, e.g., T. A. Kuhr, E. K. Sanchez, M. Skowronski, W. M. Vetter and M. Dudley, J. Appl. Phys. 89, 4625 (2001) (2001); and Y. I. Khlebnikov, R. V. Drachev, C. A. Rhodes, D. I. Cherednichenko, I. I. Khlebnikov and T. S. Sudarshan, Mat. Res. Soc. Proc. Vol. 640, p. H5.1.1 (MRS 2001), both articles being incorporated herein by reference in their entireties). In particular, poor seed attachment may cause voiding to occur between the seed and seed holder or may leave the back surface of the seed inadequately protected, allowing AlN material to sublime from that surface. For AlN crystal growth, crucible abnormalities, such as wall porosity or a seed mounting platform in which voids are present or can form, may also be a cause of voiding.
A typical planar defect 10 is shown schematically in FIG. 1. In some cases the shape of the planar defects is not perfectly hexagonal but modified or distorted and even triangular depending on the tilt between the planar void and the c-plane {0001} of AlN. In addition, there is typically a small-angle grain boundary 20 in the trail of the planar defect as shown in the schematic diagram, the origin of which is discussed below. The planar defect has a height h1, and leaves a planar defect trail of length h2 that extends back to the origin of the planar defect, typically the back of the seed crystal.
FIGS. 2a and 2b show optical microscopy images of a 2-inch diameter, c-face (i.e., c-axis oriented parallel to the surface normal of the wafer) AlN substrate taken after fine mechanical polishing. The right-side image (FIG. 2b) represents the same location as in the left-side image (FIG. 2a) taken in cross-sectioned analyzer-polarizer (AP) mode. The planar defect dimensions vary from 0.1 to 2 mm in width and up to 0.5 mm in depth, although they generally tend to be thinner (˜0.1 mm). However, the base of the planar defect is typically misoriented with respect to the overall crystal (typically a small rotation about the c-axis), and thus there is a boundary between the original crystal and the slightly misoriented material that is below the planar defect. This boundary is defined by dislocations that account for the misorientation of the material below the planar defect.
Causes of Planar Defects
If the AlN seed is poorly attached in a way that allows material in the back of the seed to move under the temperature gradient, then this material movement may cause voids to “enter” the seed. This effect is due to the fact that every void has a small but defined axial gradient that drives material to be evaporated and then re-condensed within the void. The voids entering the AlN bulk material form well-defined hexagonal-prism shapes, probably because of the anisotropy in surface energy formation.
Migration of the Planar Defect in a Thermal Gradient and Resulting Degradation of Crystal
Referring to FIGS. 3a and 3b, growth inside planar defects has been documented. The growth facet in FIG. 3b is pronounced, indicating faceted growth mode within the planar defect. Faceted growth mode usually results in a high-quality crystal. It can be expected, therefore, that the material quality within the planar defect is high and may be dislocation-free.
As the crystal grows, the planar defects effectively migrate toward the growth interface due to the axial temperature gradient within the void. Planar defects travel from the seed toward the growth interface because of the axial gradient across the planar height. As a result of this movement, the planar defects may leave “trails” (or imprints) of grain boundaries with very small misorientation angles. These small-angle grain boundaries are pronounced and shaped according to the planar defect symmetry. An example of this is shown in FIG. 4 and discussed below.
According to the traditional Read's model of low-angle grain boundaries, a boundary typically contains pure edge dislocations lying in the plane of the boundary. Therefore, after etching, the boundary is expected to exhibit a number of separated etch pits. The greater the distance between the pits, the smaller will be the misorientation angle. The grain boundary angle may be found using Frank's formula:
                                          b            D                    =                      2            ⁢                                                  ⁢            sin            ⁢                                                  ⁢                          (                              θ                2                            )                                      ,                            (        1        )            where D is the distance between dislocations (etch pits), b is the Burgers vector of dislocation, and θ is the misorientation angle. In FIG. 4, the closest distance between the etch pits is ˜12 micrometers (μm), and the Burgers vector for pure edge dislocation perpendicular to the {0001} planes is equal to the “a” lattice constant, i.e. 0.3111 nm. Therefore, the azimuthal misorientation angle of the planar defect walls is expected to be about 0.0004° (or 1.44 arcsec).
Thus, in addition to the problems caused by the physical presence of planar defects, the formation and motion of planar defects in the crystal during growth may also degrade the overall crystal quality. This degradation results because of the slight misorientation between the planar defect body and the AlN bulk material. As the planar defect moves through the crystal, it leaves behind a grain boundary, as shown in FIG. 1. These grain boundaries typically show misorientation of about 2 arcseconds for individual planar defects. However, if the density of the planar defects is high, each of these randomly misoriented grain boundaries can add up and result in much higher “effective” misorientation and, as a result, much lower crystal quality. An alternative way to look at the degradation of crystal quality is to consider the increase in threading dislocation density due to the planar defects. As one may calculate from the micrograph shown in FIG. 4, each planar defect may create over 104 dislocations/cm2 in its wake.
Problems with Surface Preparation Due to Planar Defects
Planar defects may affect preparation and polishing of AlN wafers. The sharp edges of planar defects intersecting the AlN sample surface may chip off and cause scratching. In addition, planar defects—being related to the small-angle grain boundaries (SAGB)—may result in surface roughening (topography) during chemical-mechanical polishing (CMP) treatment.
FIG. 5 shows the surface and the bulk depth of AlN containing planar defects and LAGB, where the images are obtained at the same location. It is clear that the planar defects and the SAGB cause surface roughening which, in turn, affects the epitaxial growth.
Problems with Optical Transparency and Thermal Conductivity
Planar defects may have a negative impact on the optical-transmission properties of AlN wafers because they scatter light due to the introduction of additional interfaces within the crystal, which separate regions with different refractive indices. In addition, while AlN substrates are attractive because of their high thermal conductivity (which can exceed 280 W/m-K at room temperature), planar defects may cause the thermal conductivity to diminish in a location directly above the planar defect because of the extra interfaces that are inserted at the planar defect boundaries as well as the thermal resistance of the volume of the planar defect itself. This local increase of the thermal resistance of the AlN substrate may reduce the usefulness of the AlN substrates for applications that require high power dissipation, e.g., high-power RF amplifiers and high-power, high-brightness LEDs and laser diodes.
Limitations of Existing Methods
As described in the '660 application, the production of large-diameter (i.e., greater than 20 mm) AlN crystals typically requires seeded growth. However, as discussed below, the seed holder and seed mounting technique on the holder are primary sources of planar defects in the AlN boules that are produced. The '660 application discloses a method for AlN seed attachment and subsequent crystal growth. Referring to FIG. 6, an AlN ceramic-based, high-temperature adhesive bonds the AlN seed to the holder plate and at the same time protects the back of the AlN seed from sublimation. In particular, an AlN seed 100 is mounted onto a holder plate 130 using an AlN-based adhesive 140. The AlN ceramic adhesive may contain at least 75% AlN ceramic and silicate solution that provides adhesive properties. One suitable example of such an adhesive is Ceramabond-865 available from Aremco Products, Inc.
In a particular version, the AlN seed is mounted using the following procedure:                (1) AlN adhesive is mixed and applied to the holder plate using a brush to a thickness not exceeding about 0.2 mm;        (2) The AlN seed is placed on the adhesive; and then        (3) The holder plate along with the seed is placed in a vacuum chamber for about 12 hours and then heated up to 95° C. for about 2 hours.        
This approach has proven successful in providing high-quality, large-diameter AlN crystal boules. However, planar defects as shown in FIG. 2 will form. This problem is caused by the voids left behind as the silicate solution is either evaporated or absorbed by the AlN seed crystal or by Al escaping through the seed holder.
An alternative method for AlN seed attachment and subsequent crystal growth described in the '660 application involves mounting the AlN seed on a thin foil of Al on the holder plate. The Al is melted as the temperature of the furnace is raised above 660° C. (the melting point of Al), thereby wetting the back of the seed and the holder plate. As the temperature is raised further, the Al reacts with N2 in the furnace to form AlN, which secures the seed to the holder plate. This technique may require that the AlN seed be held in place (either by gravity or mechanically) until a sufficient amount of the Al has reacted to form AlN, after which no further mechanical support is needed.
This technique, too, results in planar defects. The Al foil may melt and ball up, leaving empty spaces between agglomerations of liquid Al. The agglomerated liquid-Al metal may then react to form a nitride, leaving empty spaces between the seed and the seed holder. These empty spaces, in turn, can lead to planar defects once crystal growth is initiated on the seed crystal. The interaction between the AlN seed and the seed holder may also contribute to defects. Typically some amount of diffusion of either Al or N (or both) into the seed holder occurs at the temperatures used for crystal growth. For instance, a tungsten (W) seed holder may absorb both Al and N at the growth temperature, which can result in planar defects forming in the seed crystal and in the resulting boule grown from the seed crystal. In addition, the seed holder may have a thermal expansion coefficient different from that of the AlN crystal, which may cause defects in the seeded crystal or may induce voids to open up at the seed crystal/seed holder interface, resulting in planar defects during subsequent boule growth.
Another way to attach the seed crystal to the seed holder is to run a heat cycle under conditions whereby the seed is held onto the seed backing (e.g., by placing the seed crystal under an appropriate mass that holds the crystal down during this process), and heating the crystal up to a temperature above 1800° C. (and preferably above 2000° C.) to allow the seed to thermally, chemically and/or mechanically bond to the seed holder material. This approach is referred to herein as sinter bonding. The sintering process may, however, be difficult to control such that good bonding occurs without damaging the seed. In addition, it may be difficult to avoid leaving some space between the seed crystal and the seed holder. This space may be filled during processing with AlN that mostly comes from the seed crystal (even when vapors of Al and N2 are supplied by having an AlN ceramic present in the crucible during the sintering process), and this AlN may induce planar defects to form in the seed crystal that may propagate into the single-crystal boule grown on the seed crystal.